An intelligent search-based methodology for selection of sample points for form error estimation.

Abstract

Efficient part feature verification through CMM requires prudent sampling of data points. This dissertation presents an adaptive sampling procedure, which uses manufacturing error patterns and optimization search methods for reducing sample size, while maintaining high accuracy. The methodology is demonstrated with straightness and flatness evaluation.

Two manufacturing processes, end and face milling are used to produce plates. Respective surface errors are quantified and previous models are validated. Sampling begins with a necessary number of initial points guided by the geometry and error profiles of the object surface. The least squares method is applied to compute a tolerance zone. Next points are sampled based on search methods with suitable intensification and diversification, looking for improvements in the zone. The final value is compared with that obtained for a population sample in terms of the absolute % error. For straightness estimation, region-elimination search is used. For flatness determination, tabu search and a hybrid search are employed and their performance is compared. The hybrid search developed is a combination of coordinate search, Hooke-Jeeves search and tabu search. Experiments are conducted to investigate the effect of different factors on the sample size and % error.

Comparison with other sampling methods reveals that the present approach is more efficient and reliable. The research is expected to lead to improved solutions to inspection problems faced by industries.